Double checking: VHS workflow settings?

[licehead]

Been outta the game for a few years of digitizing vhs tapes and want to make sure I'm still on the right path.

The goal is to archive home videos and also put the final encoded versions on to USB sticks for various family members or a shared google drive

Capture setup: JVC HR-S7900 -> Cypress AVT-8120 -> ATI 600 USB clone -> VirtualDub w/ Huffyuv

Capture setup: JVC HR-S7900 -> Cypress AVT-8120 -> ATI 600 USB clone -> VirtualDub w/ Huffyuv

1. Capture - JVC settings are Video Cal - Off, Picture Control - Normal, R3 - off, Video Stabilizer -off
In VirtualDub crop edge noise to use the histogram to set levels using brightness and contrast controls. Set cropping back to 0 before capture.
2. This is where I'm slightly confused on recent threads I've read. Crop 720x480 video to 704x480. Then mask overscan noise at bottom of screen. I generally just do slight color correction with ColorMil with no other filters unless really needed.
3. Do final encode to h264 mp4 in hybrid while dinterlaing with QTGMC and resize to 640x480. Or is it worth resizing to 1440x1080 using stepped resize?

Just making sure I understand the cropping to 704 and wondering if the final resizing is worth it? Thanks

[aramkolt]

I think the 704 part is all that is actually used for the aspect ratio preservation. I've seen posts that will say that the 8 pixels on either side are not part of the 4:3 ratio, so it makes sense not to capture the black bars and do 704 as the capture resolution and ideally you'll not see the black bars in the capture file.

As for the masking of the lower bit of head switching noise, masking is the technically correct way to do it, but I will also say I don't see too many finished captures on YouTube where they keep or mask the bottom bit, seems they usually crop it for probably a slight change in aspect ratio. If the 16 pixels on the sides actually were part of the 4:3 aspect ratio, I think you'd get pretty close to preserving the 4:3 ratio by cropping those 16 pixels and maybe the 8 or 10 pixels on the bottom of noise - though you'd be left with a file that is not 480 pixels high if that matters to the end application.

Would be interested to hear how others do it as well. I know there are other forum posts, but they don't really seem to agree a lot of the times haha.